CC BY
45
S. Luraghi
INSTRUMENT AND CAUSE IN THE INDO-EUROPEAN LANGUAGES AND IN PROTO-INDO-EUROPEAN
В древних индоевропейских языках семантическая роль орудия в предложении обычно выражено формой инструменталиса (или падежа, его замещающего - аблатива в латыни или датива в древнегреческом); при этом семантическая роль причины (cause) действия выражается различными способами - как падежными формами, так и предложными конструкциями. Сопоставление данных различных индоевропейских языков позволяет реконструировать для праиндоевропейского уровня различные способы выражения причины действия - как с помощью падежных форм инструменталиса и аблатива, так и сочетаним их с наречием *pro, которое, скорее всего, уже изначально выполняло функции предлога.
Предполагается, что падежные формы и наречия-предлоги изначально имели пространственные значения, а абстрактные значения возникли в результате метафорического переноса. В случае аблатива с предлогом *pro в конструкциях причины можно проследить некоторые общие принципы семантического расширения (от источника/начала к причине: причина возникновения ситуации понимается как ее источник; от места к причине: причина понимается как ограничение ситуации). Традиционно рассматривают инструменталис как первоначальный комитатив. Расшитение функций комитатива до инструменталиса основывается на хорошо известной «метафоре спутника», при этом более раннее локативное значение праиндоевропейского инструмен-талиса восстанавливается с трудом.
Ключевые слова: инструмент, причина, падежи, предлоги, семантические роли.
0. Introduction
In this paper I compare adverbial noun phrases with the functions instrument and cause in ancient IE languages, and propose a reconstruction of the encoding of these semantic roles in PIE. While the reconstruction is straightforward for the instrument role given the existence of a dedicated case, the instrumental, the situation is more varied for the semantic role cause. As we will see further on, beside the fact that there is no specific causal case,1 evidence from the IE
1 Tocharian b is the only IE language that has a case called 'causal'. The causal case is one of secondary cases, formed by agglutination of postposed particles. Notably, Tocharian b does not have an instrumental case (while
languages allows for the reconstruction of different markers, both plain cases (instrumental, ablative), and adpositional phrases.
Adopting a localistic view of the meaning of cases and ad-positions, I also propose a reconstruction of the semantic development of individual morphemes when denoting the two semantic roles in question, and discuss the underlying metaphors that led from a spatial to more abstract meanings. The paper is organized as follows. In sec. 2, I discuss the semantic properties of the role instrument, and show how it is encoded in the ancient IE languages. I also discuss the coding of neighboring semantic roles, and other functions of the instrumental case. Sec. 3 is devoted to the semantic role cause, and its coding in the ancient IE languages. I argue that the variety of cause expressions that can possibly be reconstructed reflects the conceptual complexity of the notion of cause. Sec. 4 contains the conclusion.
1. The semantic role instrument
Prototypical features of the instrument role are inanimacy and manipulation. Moreover, the occurrence of an instrument participant implies the presence of an agent. Therefore, instrument normally occurs only in controlled situations: this constraint allows distinguishing between instruments and causes, as discussed in sec. 2.
The PIE instrumental case has reflexes in many IE languages. In several modern IE languages, as well as in other non-IE languages of Europe (for example Turkish), instrument is encoded in the same way as comitative. This can be seen in prepositional constructions of Romance and Germanic languages. Languages like Russian, which retain an instrumental case, have a 'reinforced' comitative, with the instrumental accompanied by a preposition. The cognitive-semantic link between comitative and instrument has been highlighted in Lakoff & Johnson (1980), who explain this frequent polysemy as an ontological metaphor: AN INSTRUMENT IS A COMPANION. This metaphor is found in genetically unrelated languages, but is more common in European languages than in other geographic areas, as demonstrated by Stolz et al. (2006). Also frequent in languages of various affiliations is the encoding of the comitative through the instrumental case plus an adposition, as in Slavic languages, or an adverb, as in English together with (here, together strengthens the
Tocharian a, which has the instrumental, does not have a causal case). As secondary cases in Tocharian do not reflect the PIE case system, I have not included Tocharian in the present discussion.
preposition with, but is not grammaticalized hence not obligatory), as noted in Stolz (1998).2 From a diachronic point of view, comitative markers often extend to instrument; this semantic extension may be followed by additional marking of comitative by means of an adposition or an adverb, which then sometimes extends again to instrument (Luraghi 2001). This dual evolution can be easily illustrated with data from the early IE languages, as we will see in sec. 1.1.
1.1. The PIE instrumental
The instrumental case of PIE underwent the semantic evolution described in the previous section, based on the Companion metaphor. Indeed, the primary meaning of the IE instrumental was comitative. Delbrück (1867: 50) writes about Sanskrit: "Wenn die indische grammatik den instrumentalis den kategorien werkzeug und urheber unterordnet, so bezeichnet sie damit dasjenige gebiet, auf welchem der instrumentalis am häufigsten gebraucht wird. Der grundbegriff des instrumentalis aber ist der des Zusammenseins." Delbrück also mentions 'bridging' contexts that could enable the extension from comitative to instrument.
More complex is the reconstruction of a possible original spatial meaning for the instrumental, which, in a localistic perspective, should precede the comitative meaning. The IE languages that preserve an instrumental case also attest to spatial usages, but a connection with comitative is problematic. In general, spatial functions of the instrumental in IE languages point toward a perlative meaning, with no special association with animate nouns, as shown in (1) and (2).
Vedic Sanskrit
(1) a sahasrampathibhir indra raya ... yahi
'Kingly Indra, come on a thousand paths.' (Rv. I 1811).
(2) diva yanti maruto bhumyagnir ayam vato antariksena yati
'The Maruts move in heaven, on earth moves Agni; through the mid-firmament the Wind approaches.' (Rv. I 16114).
It is hard to see how the spatial meaning attested in occurrences such as (1) and (2) may have given origin to the sociative meaning.
2 Stolz cites among others the example of Alyawarra, an Australian language, which has an instrumental suffix -ila; the comitative is formed from the instrumental with the addition of another suffix: -ila-linga.
As the comitative is a semantic role prototypically assigned to human participants, one would expect that comitative morphemes would derive from markers of spatial relations used with human landmarks. Remarkably, this process can be observed in languages such as French or Ancient Greek. French avec comes from Latin apud, which most often indicated proximity to humans (Luraghi 2010a), while Ancient Greek meta+gen was used in Homer with the meaning of 'among' only with human landmarks (see Chantraine 1953: 119, Luraghi 2003: 245-246). Nevertheless, as discussed in Luraghi (2014: 109-110), in these languages prepositions that express comitative have not retained their original spatial meaning: in French, avec does not mean 'at', and in Classic Greek meta+gen does not mean 'among'. The infrequency of a merger between comitative and locative is also noted in Stolz et al. (2006: 140-147, 361), who note that "Comitative escape being classified as just another instance of a concrete local case."
In sum, the prepositions that express comitative in IE languages may have a spatial meaning, but this meaning is most likely secondary with respect to the meaning of comitative, rather than be its origin. On the other hand, the original spatial meaning does not seem to be preserved. For example, Latin cum 'with' did not preserve any trace of an earlier spatial meaning, and only comparative evidence, which connects Latin cum with Oscan ku '(near)-by', allows one to reconstruct it (Leumann, Hofmann & Szantyr 1965: 260). Similarly, German mit 'with' is etymologically related with the word Mitte 'middle', but does not share its meaning (see further Luraghi 2014: 109). Thus, it seems very doubtful that one can assume a specific spatial meaning for the PIE instrumental case.
1.2. IE languages with an instrumental case
In IE languages that preserve the instrumental case, this is normally used to encode instrument. Examples are (3)-(7).
Sanskrit
(3) abhi jahi raks&sah p&rvatena
'Hit the Raksasa with the stone bat.' (Rv. Vii 10419).
Hittite
(4) s = an ispanti nakkit dahhun
'I took it (sc. the town) at night with an assault.' (18 StBoT 47-48).
Armenian
(5) arar zawrut'iwn bazkaw iwrov
'He performed a miracle by his own hand.' (Luke 1:51).3
Old Church Slavonic
(6) нъ посълавыи мл кръститъ въ водп>
'But he sent me to baptize with water.' (John 1:33).
Lithuanian
(7) Tuomet Viespats lydino ant Sodomos ir Gomoros siera ir ugnimi... 'Then the Lord rained on Sodom and Gomorrah with brimstone and fire...' (Gen. 19.24).
It needs to be remarked that these languages do not agree completely in preserving all other functions of the instrumental: in particular, in most languages the comitative role is encoded by the instrumental case accompanied by an adposition. This is true of Slavic languages, while in early Vedic the instrumental case may occasionally encode the comitative role even without further specification, but most often it occurs with an adverb (later postposition), such as saha, sakam, sumad, smad 'with', or the comitative relation may be indicated by the preverb sam. In Classical Armenian, the comitative is encoded through adpositional phrases with the instrumental or with other cases (dative or locative) such as and 'with'+loc (+dat in case of pronouns), handerj+instr 'together with'. In Hittite, the adverb/adposition kattan 'with' occurs with the genitive or with the dative/locative in order to encode comitative. Thus, the comparative data point toward an initial comitative meaning of the instrumental case, along with an ongoing tendency to reinforce the comitative through dedicated adpositions or adverbs, in line with the process described in sec. 1.1.
Another possible function of the instrumental in the IE languages is to indicate the agent with passive verbs. This function is reconstructed essentially based on Indo-Iranian, as other languages that preserve an instrumental case do not use it consistently to encode the agent. For example in Armenian and Baltic languages, animate agents of passive verbs are encoded through prepositional phrases (see Luraghi 1986). In Slavic languages, the instrumental encodes human agent of passive verbs to different extents: in
3 From Meillet (1936: 96), see further Jensen (1959: 179-180) on the Armenian instrumental. Both Meillet and Jensen note that the comitative tends to be reinforced by prepositions, similar to Latin and Slavic languages.
particular, in Old Church Slavonic one can find either the instrumental, or, often, prepositional phrases with otu+gen as in (8), and the same type of prepositional phrase was also a possible alternative to the instrumental of agent in Old Russian (see Brauer 1952 on OCS).4
Old Church Slavonic
(8) бждете ненавидими отъ всъхъ имене моего ради. 'You will be hated by all for my name.' (Luke 21.17).
In Hittite, the instrumental of agent with passive verbs, especially participles, also occurs as shown in (9), but notably only after the Old Hittite period: no agent phrases with passive verbs occur in Old Hittite original texts.5
Hittite
(9) GISTUKUL.HI.A-i.s=wa=tta siunit piyantes
'Weapons have been given to you by the god.' (KBo 22.6 + KUB 48.98 i 24-25).
Hoffner & Melchert (2008: 269) assume this to be a coincidence, neverthless, there are reasons that indicate that this might not be the case. In the first place, according to Starke (1977), nouns with human referents never occur in the instrumental (or ablative) case in Old Hittite. In the second place, low grammaticalization of passive voice in the early IE languages makes it doubtful that an agented passive construction already existed in PIE. Indeed, while some scholars think that the evidence unequivocally points to the reconstruction of the instrumental of agent with human nouns in PIE (e.g. Jamison 1979), others believe that the construction is the outcome of parallel, yet unrelated developments in the IE languages (see Luraghi 1986, Hettrich 1990, Strunk 1991), and this on account of our knowledge about possible reconstruction of the PIE voice system, which seems to rule out the existence of a grammaticalized passive (see further Luraghi 2010b with the references therein).
4 Brauer (1952) notes that the distribution of the instrumental of agent and adpositional phrases in OCS correlates with tense and aspect, and that reflexive passives only occur with prepositional agent phrases. He concludes that the existence of different agent expressions in OCS and in general in Slavic languages indicates the recent rise of agented passives. For a similar remark on Homeric Greek see Luraghi (2000).
5 Some ablatives of agent, also post-Old Hittite, may be taken as replacing the instrumental case, which increasingly merged with the ablative after the Old Hittite period (see Hoffner & Melchert 2008: 267).
Rather, one might wonder whether, in the languages in which the instrumental has experienced the extension to agent, this might have brought about the reinforcement of the comitative instrumental through adpositions. In fact a polysemy that includes agent and comitative is infrequent (Stolz 2001, Stolz et al. 2006: 277), most likely because it would bring about a high degree of ambiguity, as both semantic roles are typically assigned to humans, and can occur in the same types of situations. The hypothesis would therefore appear plausible; note, however, that the reinforcement of the comitative instrumental is a phenomenon that goes beyond languages that show polysemy of agent and instrument. Also in the light of data from other languages (for example, those collected in Stolz et al. 2006), it seems that there is a general tendency for comitative markers, after having extended to instrument, to become again distinct through the addition of a second marker (see the discussion in sec. 1.3).
Finally, the instrumental case is also attested in West Germanic languages: despite its marginal existence, it still occurs without prepositions in some passages (see Delbrück 1907: 156-167). Most often, the instrumental is strengthened, also in the role instrument, by prepositions that also encode comitative.
1.3. The outcome of the instrumental in Latin and Greek
Latin and Greek do not preserve a case called instrumental in grammars. Nevertheless, both languages, in different ways, feature cases used mostly for encoding the instrument role.
In Latin, the plain ablative has, in most cases, the function of an instrumental: spatial functions of the ablative outside prepositional phrases are lexically restricted (Luraghi 2010a). For the encoding of comitative, Latin shows the same development as the Slavic languages, relying on preposition phrases with the ablative. The plain ablative as comitative is used only marginally, with collective nouns that have a low degree of individuation, and can also be interpreted as instruments, although not prototypical (so-called ablativus militaris):
Latin
(10) Illi equitatu atque essedis ad flumen progressi ex loco superiore nostros prohibere et proelium committere coeperunt. 'They advanced with cavalry and chariots from the heights to the river and began to hinder our troops and engage in fighting.' (Caes. BG 5.9.3).
On the other hand, in Latin polysemy of agent and instrument does not involve prototypical agents, but only holds for inanimate entities, which partly share the features of instruments (see Luraghi 1986, 2010a).
Latin exemplifies what we might call the 'comitative cycle': there are no apparent reasons why the plain ablative could not indicate prototypical comitative in Latin, as it is not used to encode human passive agents, as the instrumental of Sanskrit or Russian, and in general has no significant uses with human nouns, as is the case instead of the Greek dative, discussed further ahead. However, since the time of the earliest sources comitative encoding is accomplished through cum+abl: in other words, with a 'reinforced' instrumental. In Late Latin, with the case system fading away, the instrumental ablative was replaced by cum+abl. Today, the Romance languages feature the polysemy of comitative and instrument, although comitative expressions can occasionally be reinforced as in Italian insieme con 'together with'.
In Ancient Greek the semantic role instrument is encoded through the plain dative. The Greek dative is a composite case, resulting from syncretism of the IE dative, locative and instrumental. Similar to the Latin ablative, the Greek dative is subject to lexical restrictions when encoding spatial relations, so that, as also noted in De la Villa (1989), the plain dative of inanimate nouns mostly continues the PIE instrumental. In this way, Greek relies on an important feature of the IE dative 'proper', that is, its strong connection with semantic roles typically assigned to human entities, such as recipient or beneficiary. Based on this feature, the plain dative shows a two-fold distinction between nouns with human referents, which take typical functions of the IE dative, and nouns with inanimate referents, which typically indicate instrument (or other functions often associated with the instrumental, such as manner). This limits the possibility of continuing the comitative meaning of the PIE dative: as the comitative is also a semantic role typically assigned to human entities, dative encoding would create a high level of ambiguity. And indeed, along with agent-comitative polysemy, comitative-beneficiary polysemy is extremely rare, almost inexistent cross-linguistically, as argued in Stolz et al. (2006). In Homer and often in Ionic, comitative is encoded by sMn+dat, that is, again, a reinforced instrumental, while in Classical Greek one mostly finds a new type of prepositional phrase, meta+gen. Thus, Classical Greek, especially Attic, has a distinct pattern with respect to most IE
languages, which feature either the Companion metaphor or a reinforced instrumental. In Byzantine Greek, when the dative case disappeared, the Companion metaphor was again responsible for extension of meta+gen/acc (and later me+acc) to instrument (Luraghi 2005). The comitative dative survived in Classical Greek in some idiomatic expressions, as shown in (11) (see Crespo 1994).
Greek
(11) nsvrs Ss elafiov, Kai piav rovrmv avroig dvSpdaiv
'They took five (ships), one of them with the crew.' (Th. 4.14.1).
2. The semantic role Cause
With respect to instrument, cause is conceptually more complex. While there is a natural class of instruments (weapons, tools, vehicles, body parts), and while the prototypical features of instruments are easily detectable, there is no natural class of causes, as the entities that can occur in this semantic role are quite diverse, and include natural phenomena, emotions, concrete and abstract inanimate entities, human beings, and, very often, situations. Moreover, while the role instrument implies the occurrence of an agent acting voluntarily and exerting control over the situation, the role cause has no implications regarding the possible occurrence or non-occurrence of an agent. Let us consider examples (12)-(15).
(12) The asphalt broke due to the high temperatures.
(13) The child trembled out of fear.
(14) Mary passed the exam thanks to John / because John helped her.
(15) I left because I had other commitments.
In (12) and (13) we find uncontrolled situations, while in (15) the agent acts intentionally (this type of cause, which motivates an agent to act consciously, may also be called reason). In (14) the cause does not per se bring about the situation, but enables the agent in bringing it about. Note that in this example the enabling cause is animate.
Even when they are inanimate, possible causes are distinct from prototypical instruments, as they are not necessarily concrete entities, and are often non-manipulated. Based on the feature of control, the roles instrument and cause are partly complementary, as instruments occur necessarily in controlled situations, while causes can occur both in controlled and in uncontrolled ones. Thus, with uncontrolled situations it is not necessary to keep the two semantic roles distinct through different coding means.
The notion of cause has been studied by Turner (1987) from a cognitive linguistics point of view. Turner has shown that, using a folk model of causation, the cause of a situation is not conceived as a necessary and sufficient condition, as in logic, but rather as its origin. The ontological metaphor CAUSES ARE ORIGINS has been discussed in Nikiforidou (1991: 175-176). Following this metaphor, one can expect spatial expressions that indicate origin to extend to the encoding of cause. Indeed, this is the case for the IE ablative.
Although the conceptual contact between cause and origin provides an important source for cause expressions, cross-linguistic evidence points toward a wider variety in the encoding of cause than in the encoding of instrument. As remarked by Dirven (1995) and Radden (1985), in English, for example, virtually all prepositions show semantic extensions that bring them to indicate cause. Consequently, we may expect to encounter more difficulties in the reconstruction of PIE cause expressions than in the reconstruction of instrument expressions.
2.1. Instrumental of cause
Languages that preserve the instrumental can use it for the coding of cause, especially in uncontrolled situations. Remarkably, even the moribund instrumental of West Germanic languages, which was often reinforced with a preposition in instrument expressions, helps reconstruct the PIE instrumental of cause. Examples are (16)-(18). Sanskrit
(16) jarasa marate patir
'The bride dies from old age.' (Rv. X 8611).
Hittite
(17) DUMU.LU.ULUmeS DINGIRmeS-s = a kistantit harkiyanzi 'The sons of men and the gods are dying from hunger.'
(KUB 17:10 i 18:30).
Old High German
(18) thu hungiru nirstirbist
'You do not die from hunger.' (Otfr. 22/02/22).
The instrumental of cause is also well attested with controlled situations: in such occurrences, the distinction between cause and instrument may be ensured by the occurrence of nouns that indicate non-manipulated entities (nouns of emotions, action nouns, etc.), as in (19) and (20).
Sanskrit
(19) sä bhisä ni lilye
'He hid for fear.' (Satapatha Brahmana 1.2.3.1);
(20) avidyaya-iva tad ahuh
'Indeed, they say this out of ignorance.'
(Aitareya Brahmana 11.01.10).
The Latin ablative, which continues both the PIE ablative and the PIE instrumental, can encode cause, but due to case syncretism it is impossible to understand whether this usage is inherited from either case.
In Ancient Greek, in which the instrumental underwent syncretism, but not with the ablative, the usage of the plain dative offers evidence for the instrumental of cause. Similar to the Sanskrit instrumental, the Greek dative can indicate cause both in uncontrolled and in controlled situations, as shown in (21)-(23).
Greek
(21) (pQövo? Karapapysmv
'Mad with envy.' (Hdt. 8.125.1).
(22) öaoi ysvsasi каг трофагд sv nenaidevvmi
'For those who, by birth and nurture, are easily educated.'
(Pl. Laws 920a8).
(23) iSsiv ¡¡sv yap щю^ЦУ ещ щу nozs 'Орщюд ysvofievtyv kvkvov ßiov aipovjievtyv, ¡¡iosi zov yvvaiKsiov ysvovq
'He said he saw the soul that had been of Orpheus choosing the life of the swan, out of hatred for the female gender.' (Pl. Rep. 620a).
2.2. Ablative of cause
The ablative case is less well preserved than the instrumental in IE languages: only in Indo-Iranian, Anatolian and Latin have a separate ablative case, but the Latin ablative has merged with the instrumental, as already remarked above. Thus, even though the ablative of cause is well attested in Latin, strictly speaking, it cannot be used for the reconstruction of PIE, as its occurrence may be a reflex of the PIE instrumental. In Sanskrit occurrences, as pointed out by Delbrück (1888: 114-115), the notion of origin is always present as shown in (24), and the same is true for the only example of an ablative of cause from an Old Hittite original, quoted in (25) (see further Hoffner & Melchert 2008: 265).
Sanskrit
(24) ma nas tasmad enaso deva ririsah
'Oh gods, let us not suffer for this sin.' (Rv. Vii 895).
Hittite
(25) nu=za halluwayanza mekki nahhantes esten
'Be very cautious because of a quarrel (i.e.: try to avoid it).'
(KUB 13.4 iii 43).
Languages such as Slavic, Baltic and Greek, in which the ablative merged with the genitive, have various types of adpositional phrases in cause expressions, and the same holds for Germanic. Part of these adpositions also express origin, thus showing the same pattern of semantic development as the PIE ablative.
2.3. An adpositional construction
Dunkel (1990) pointed out that, among various adpositional phrases attested in ancient IE languages, one is likely to reflect a PIE construction. Consider examples (26)-(29).
Greek
(26) nspi yap Sis рц piv A/aioi apyalsov про фовою sXmp Sq'ioiai Xinoisv. 'For he feared that the Achaeans in should leave him to be a prey to the foemen out of their own fear.' (1l. 17.666-667);
Hittite
(27) nu=mu E-FA inani peran pittuliyas E-ir kisat nu=mu pittuliyai peran istanzas=mis tamatta pedi zappiskizzi
'Because of illness my house became a house of fear; for the pain my soul drips elsewhere (ie: 'my life is gone).' (KUB 30.10 14);
Gothic
(28) ip bipe gabadran ist barn, ni panaseips ni gaman pizos aglons fadra fahedai
'But when her baby is born [a woman] forgets the anguish because of her joy.' (John 16:21);
Latin
(29) ut prae timore in genua in undas concidit
'Out of fear she fell into the waves upon her knees.' (Rud. 174)
In these examples we find a number of etymologically related adpositions: Greek про, Hittite peran, Gothic fadra and Latin prae. Their spatial meaning is 'in front', 'before'; they can all be used in cause expressions with different frequency and partly different
distribution depending in the individual language. In Greek the use of npo for the encoding of cause is exceptional, and virtually limited to this passage. In Hittite, peran, which occurs twice in (27) is frequently for the encoding of cause, at all language stages. Latin prae most often denotes the causa impedientis. Finally, the Gothic example reflects a generalized use in numerous Germanic languages for the expression of this semantic role (cf. for example German vor).
Dunkel also mentions some occurrences from Indo-Iranian languages, where cognate adpositions can express cause, although marginally. Also on account of the marginality of such occurrences, Dunkel concludes that a construction with *pro must be reconstructed, which already indicated cause in PIE. In other words, the shift from 'before' to 'because of' had already taken place at the stage of the proto-language.
Semantic extension from location in front of a landmark to cause can be explained as a consequence of the conceptualization of a cause as something that manifests itself in front of a participants, and conditions his/her reactions. Following this type of construal, the cause is conceptualized as conditioning a situation by means of its presence. Note that, according to an embodied metaphor, being positioned in front of a participant corresponds to being visible. Therefore, the presence of something which is located in front cannot be ignored. The cause is thus conceptualized as an entity whose presence and visibility have an effect on the event, and can generate specific consequences. This metaphor is also active in English, in expressions such as 'in view of', as in (30):
(30) In view of a possible rise in the cost of oil, truck drivers went on strike.
The English expression still has a strong lexical motivation, and can be used only in certain contexts, but clarifies the way in which the PIE adverb must have acquired its causal meaning.
In the case of adpositions or adverbs that derive from *pro, we cannot only reconstruct a certain metaphor, as we can do in order to explain the occurrence of source adpositions to express cause in languages in which the ablative case disappeared. More interestingly, in this case we can also reconstruct a specific lexeme reflected by various forms in the IE languages, which could already encode cause in the proto-language, in addition to plain cases, i.e. to the instrumental and the ablative of cause.
3. Conclusion
From the discussion in the preceding sections we can draw the following conclusions. In the first place, the semantic role instrument was encoded in PIE by the instrumental case. The original meaning of this case was comitative, and the semantic extension to instrument is motivated by the similarity of these two roles at the cognitive level, which is explained by the well-studied Companion metaphor.
The semantic role cause, in its turn, did not rely on a dedicated case, and its coding was more complex than the coding of instrument. We may view the greater variety of formal means for the expression of cause as a consequence of its conceptual complexity as already noted in sec. 2. Based on comparative data, one can reconstruct at least three different ways of encoding cause in PIE:
a) the instrumental case: the extension of the instrumental from instrument to cause is possible based on the partly different distribution of the two semantic roles (instrument can only occur in controlled situations) and on the types of referents that can take the two roles (instrument is limited to inanimate and concrete entities);
b) the ablative case: the ablative case denotes source and origin; the extension leading from origin to cause is based on a metaphor by which we construe causes as origins of events;
c) the adverb/adposition *pro 'in front': in this case the cause is conceptualized as an entity whose presence has the power to constrain an event.
References
Bräuer 1952 - Bräuer H. Der persönliche Agens beim Passiv im Altbulgarischen. Eine syntaktischen Untersuchung. Abhandlungen der Akademie der Wissenschaft und der Literatur in Mainz. Wiesbaden, 1952. Chantraine 1953 - Chantraine P. Grammaire homérique II: Syntaxe. Paris: Klincksieck, 1953.
Crespo 1994 - Crespo E. L'expression de l'accompagnement en grec ancien. In B. Jacquinod ed., Cas et prépositions en grec ancien. Actes du colloque international de Saint-Etienne. Saint-Étienne: Publications de l'Université, 1994. P. 181-192. De la Villa 1989 - De la Villa J. Caractérisation fonctionelle du datif grec.
Glotta. 1989. Vol. 67. P. 20-40. Delbrück 1867 - Delbrück B. Ablativ localis instrumentais im Altindischen, Lateinischen, Griechischen und Deutschen: ein Beitrag zur vergleichenden Syntax der indogermanischen Sprachen. Berlin: Dummler, 1867.
Delbrück 1888 - Delbrück B. Altindische Syntax. Halle, 1888 (repr. 1968,
Darmstadt: Wissenschaftliche Buchgesellschaft). Delbrück 1907 - Delbrück B. Syncretismus. Ein Beitrag zur germanischen Kasuslehre. Strassburg: Trübner, 1907.
Dirven 1995 - Dirven R. The construal of cause: The case of cause prepositions // J. R. Taylor, R. E. MacLaury eds. Language and the Cognitive Construal of the World. Berlin: Mouton de Gruyter, 1995. P. 95-118.
Dunkel 1990 - Dunkel G. prae pavore, про 90ßoio // Indogermanische Forschungen. 1990. Bd. 95. S. 161-170.
Hettrich 1990 - Hettrich H. Der Agens in passivischen Sätzen altindogermanischer Sprachen. Göttingen: Vandenhoeck & Ruprecht, 1990.
Hoffner, Melchert 2008 - Hoffne H. A., Melchert H. C A Grammar of the Hittite Language. Winona Lake: Eisenbrauns, 2008.
Jamison 1979 - Jamison S. W. The case of the agent in Indo-European // Die Sprache. 1979. Bd. 25/2. P. 129-143.
Jensen 1959 - Jensen H. Altarmenische Grammatik. Heidelberg: Winter, 1959.
Lakoff, Johnson 1980 - Lakoff G., Johnson M. Metaphors we live by, Chicago: Chicago UP, 1980.
Leumann, Hofmann, Szantyr 1965 - Leumann M., Hofmann J. B., Szantyr A. Lateinische Syntax und Silistik. München: Beck, 1965.
Luraghi 1986 - Luraghi S. On the distribution of instrumental and agent markers for human and non-human agents of passive verbs in some Indo-European languages. Indogermanische Forschungen 91, 48-66.
Luraghi 2000 - Luraghi S. Spatial metaphors and agenthood in Ancient Greek // Ch. Zinko & M. Offisch, Hrsg. 125 Jahre Indogermanistik in Graz. 2000. P. 283-298.
Luraghi 2001 - Luraghi S. Some remarks on Instrument, Comitative, and Agent in Indo-European // Sprachtypologie und Universalienforschung. 2001. Bd. 54/4. P. 385-401.
Luraghi 2003 - Luraghi S. On the Meaning of Prepositions and Cases. A Study of the Expression of Semantic Roles in Ancient Greek. Amsterdan: Benjamins, 2003.
Luraghi 2005 - Luraghi S. The history of the Greek preposition metá: from polysemy to the creation of homonyms // Glotta. 2005. Vol. 81. P. 130159.
Luraghi 2010a - Luraghi S. Adverbial Phrases // Ph. Baldi and P. Cuzzolin (eds.). A New Historical Syntax of Latin. Berlin: Mouton de Gruyter, 2010. P. 19-107.
Luraghi 2010b - Luraghi S. The Extension of the transitive construction in Ancient Greek // Acta Linguistica Hafniensia. 2010. Vol. 42/1. P. 6074.
Luraghi 2014 - Luraghi S. Plotting diachronic semantic maps: the role of metaphors // S. Luraghi & H. Narrog eds. Perspectives on semantic roles. Amsterdam: Benjamins, 2014. P. 99-150.
Meillet 1936 - Meillet A. Grammaire comparée de l'arménien classique. Vienne: Imprimerie des PP. Mékhitaristes, 1936.
Nikiforidou 1991 - Nikiforidou K. The meaning of the genitive. Cognitive Linguistics. 1991. Vol. 2/2. P. 149-205.
Quirk, R., S. Greenbaum, G. Leech & J. Svartvik 1985. A Comprehensive Grammar of the English Language. London-New York: Longman.
Radden 1985 - Radden G. Spatial metaphors underlying prepositions of causality // Paprotté, W. & R. Dirven eds. The Ubiquity of Metaphor, Amsterdam: Benjamins, 1985. P. 177-207.
Starke 1977 - Starke F. Die Funktionen der dimensionalen Kasus und
Adverbien im Althethitischen. Wiesbaden: Harrassowitz, 1977. Stolz 1996 - Stolz, Th. Some instruments are really good companions -some are not. On syncretism and the typology of instrumentals and comitatives // Theoretical Linguistics. 1996. Vol. 23. P. 113-200. Stolz 1998 - Stolz Th. Komitative sind mehr als Instrumentale sind mehr als Komitative. Merkmahlhaftigkeit und Markiertheit in der Typologie der mit-Relationen // Sammelband des II. internationalen Symposions zur Natürlichkeitstheorie, Maribor: Pedagogische Fakultät der Universität, 1998. P. 135-167. Stolz 2001 - Stolz Th. Comitatives vs. instrumental vs. agents // W. Bisang, Aspects of Typology and Universals, Berlin: Akademie Verlag, 2001. P. 153-174. Stolz, Stroh, Urdze 2006 - Stolz Th., Stroh C., Urdze A. On comitatives and Related Categories. A Typological Study with Special Focus on the Languages of Europe. Berlin: Mouton de Gruyter, 2006. Strunk 1991 - Strunk K. Syntaktische Bemerkungen zum hehitischen und indogermanischen Instrumental // Историческая лингвистика и типология. Moskva: Akademija Nauk, 1991. S. 81-91. Turner 1987 - Turner M. Death is the Mother of Beauty: Mind, Metaphor, Criticism. Chicago: Chicago UP, 1987.
S. Luraghi. Instrument and Cause in the Indo-European languages and in Proto-Indo-European
While the semantic role instrument is uniformly encoded through the instrumental case (or cases that replaced it, such as the Latin ablative or the Greek dative) in ancient Indo-European (IE) languages, cause is variously encoded through plain cases or adpositional phrases. Comparison allows reconstructing a similar situation for Proto-Indo-European (PIE) as well. In particular, one can reconstruct cause expressions involving the plain instrumental, the plain ablative, and the adverb *pro, which, most likely, already had an adpositional function in the proto-language. It is assumed that cases and adverbs/adpositions originally had a spatial meaning, and that abstract meanings came about through metaphoric extension. In the case of the ablative and of *pro in cause expressions, some common patterns of semantic extension are visible (from source/origin to cause: the cause of a situation is conceived as its origin; from location to cause: a cause is conceived as an entity which constrains a situation by being placed in front of it). The instrumental case is traditionally thought as having originated as a comitative. While extension from comitative to instrument is based on the well-known 'Companion metaphor', an earlier local meaning of the PIE instrumental case cannot be reconstructed easily.
Keywords: instrument, cause, cases, adpositions, semantic roles.
